Jean-Francois Castet

Satellite and satellite subsystems reliability: Statistical data analysis and modeling

Reliability has long been recognized as a critical attribute for space systems. Unfortunately, limited on-orbit failure data and statistical analyses of satellite reliability exist in the literature. To fill this gap, we recently conducted a nonparametric analysis of satellite reliability for 1584 Earth-orbiting satellites launched between January 1990 and October 2008. In this paper, we extend our statistical analysis of satellite reliability and investigate satellite subsystems reliability. Because our dataset is censored, we make extensive use of the Kaplan–Meier estimator for calculating the reliability functions. We derive confidence intervals for the nonparametric reliability results for each subsystem and conduct parametric fits with Weibull distributions using the maximum likelihood estimation (MLE) approach. We finally conduct a comparative analysis of subsystems failure, identifying the “culprit subsystems†that drive satellite unreliability. The results here presented should prove particularly useful to the space industry for example in redesigning subsystem test and screening programs, or providing an empirical basis for redundancy allocation.

Beyond Reliability, Multi-State Failure Analysis of Satellite Subsystems: A Statistical Approach

Abstract Reliability is widely recognized as a critical design attribute for space systems. In recent articles, we conducted nonparametric analyses and Weibull fits of satellite and satellite subsystems reliability for 1584 Earth-orbiting satellites launched between January 1990 and October 2008. In this paper, we extend our investigation of failures of satellites and satellite subsystems beyond the binary concept of reliability to the analysis of their anomalies and multi-state failures. In reliability analysis, the system or subsystem under study is considered to be either in an operational or failed state; multi-state failure analysis introduces “degraded states” or partial failures, and thus provides more insights through finer resolution into the degradation behavior of an item and its progression towards complete failure. The database used for the statistical analysis in the present work identifies five states for each satellite subsystem: three degraded states, one fully operational state, and one failed state (complete failure). Because our dataset is right-censored, we calculate the nonparametric probability of transitioning between states for each satellite subsystem with the Kaplan–Meier estimator, and we derive confidence intervals for each probability of transitioning between states. We then conduct parametric Weibull fits of these probabilities using the Maximum Likelihood Estimation (MLE) approach. After validating the results, we compare the reliability versus multi-state failure analyses of three satellite subsystems: the thruster/fuel; the telemetry, tracking, and control (TTC); and the gyro/sensor/reaction wheel subsystems. The results are particularly revealing of the insights that can be gleaned from multi-state failure analysis and the deficiencies, or blind spots, of the traditional reliability analysis. In addition to the specific results provided here, which should prove particularly useful to the space industry, this work highlights the importance of conducting, beyond the traditional reliability analysis, multi-state failure analysis of any engineering system when seeking to understand its failure behavior.

Single versus mixture Weibull distributions for nonparametric satellite reliability

Long recognized as a critical design attribute for space systems, satellite reliability has not yet received the proper attention as limited on-orbit failure data and statistical analyses can be found in the technical literature. To fill this gap, we recently conducted a nonparametric analysis of satellite reliability for 1584 Earth-orbiting satellites launched between January 1990 and October 2008. In this paper, we provide an advanced parametric fit, based on mixture of Weibull distributions, and compare it with the single Weibull distribution model obtained with the Maximum Likelihood Estimation (MLE) method. We demonstrate that both parametric fits are good approximations of the nonparametric satellite reliability, but that the mixture Weibull distribution provides significant accuracy in capturing all the failure trends in the failure data, as evidenced by the analysis of the residuals and their quasi-normal dispersion.

On the concept of survivability, with application to spacecraft and space-based networks

Abstract Survivability is an important attribute and requirement for military systems. Recently, survivability has become increasingly important for public infrastructure systems as well. In this work, we bring considerations of survivability to bear on space systems. We develop a conceptual framework and quantitative analyses based on stochastic Petri nets (SPN) to characterize and compare the survivability of different space architectures. The architectures here considered are a monolith spacecraft and a space-based network. To build the stochastic Petri net models for the degradations and failures of these two architectures, we conducted statistical analyses of historical multi-state failure data of spacecraft subsystems, and we assembled these subsystems, and their SPN models, in ways to create our monolith and networked systems. Preliminary results indicate, and quantify the extent to which, a space-based network is more survivable than the monolith spacecraft with respect to on-orbit anomalies and failures. For space systems, during the design and acquisition process, different architectures are benchmarked against several metrics; we argue that if survivability is not accounted for, then the evaluation process is likely to be biased in favor of the traditional dominant design, namely the monolith spacecraft. If however in a given context, survivability is a critical requirement for a customer, the survivability framework here proposed, and the stochastic modeling capability developed, can demonstrate the extent to which a networked space architecture may better satisfy this requirement than a monolith spacecraft. These results should be of interest to operators whose space assets require high levels of survivability, especially in the light of emerging threats.

Interdependent Multi-Layer Networks: Modeling and Survivability Analysis with Applications to Space-Based Networks

This article develops a novel approach and algorithmic tools for the modeling and survivability analysis of networks with heterogeneous nodes, and examines their application to space-based networks. Space-based networks (SBNs) allow the sharing of spacecraft on-orbit resources, such as data storage, processing, and downlink. Each spacecraft in the network can have different subsystem composition and functionality, thus resulting in node heterogeneity. Most traditional survivability analyses of networks assume node homogeneity and as a result, are not suited for the analysis of SBNs. This work proposes that heterogeneous networks can be modeled as interdependent multi-layer networks, which enables their survivability analysis. The multi-layer aspect captures the breakdown of the network according to common functionalities across the different nodes, and it allows the emergence of homogeneous sub-networks, while the interdependency aspect constrains the network to capture the physical characteristics of each node. Definitions of primitives of failure propagation are devised. Formal characterization of interdependent multi-layer networks, as well as algorithmic tools for the analysis of failure propagation across the network are developed and illustrated with space applications. The SBN applications considered consist of several networked spacecraft that can tap into each others Command and Data Handling subsystem, in case of failure of its own, including the Telemetry, Tracking and Command, the Control Processor, and the Data Handling sub-subsystems. Various design insights are derived and discussed, and the capability to perform trade-space analysis with the proposed approach for various network characteristics is indicated. The select results here shown quantify the incremental survivability gains (with respect to a particular class of threats) of the SBN over the traditional monolith spacecraft. Failure of the connectivity between nodes is also examined, and the results highlight the importance of the reliability of the wireless links between spacecraft (nodes) to enable any survivability improvements for space-based networks.

Spacecraft electrical power subsystem: Failure behavior, reliability, and multi-state failure analyses

Abstract This article investigates the degradation and failure behavior of spacecraft electrical power subsystem (EPS) on orbit. First, this work provides updated statistical reliability and multi-state failure analyses of spacecraft EPS and its different constituents, namely the batteries, the power distribution, and the solar arrays. The EPS is shown to suffer from infant mortality and to be a major driver of spacecraft unreliability. Over 25% of all spacecraft failures are the result of EPS failures. As a result, satellite manufacturers may wish to pursue targeted improvement to this subsystem, either through better testing or burn-in procedures, better design or parts selection, or additional redundancy. Second, this work investigates potential differences in the EPS degradation and failure behavior for spacecraft in low earth orbits (LEO) and geosynchronous orbits (GEO). This analysis was motivated by the recognition that the power/load cycles and the space environment are significantly different in LEO and GEO, and as such, they may result in different failure behavior for the EPS in these two types of orbits. The results indicate, and quantify the extent to which, the EPS fails differently in LEO and GEO, both in terms of frequency and severity of failure events. A casual summary of the findings can be stated as follows: the EPS fails less frequently but harder (with fatal consequences to the spacecraft) in LEO than in GEO.

Survivability and Resiliency of Spacecraft and Spac e-Based Networks: a Framework for Characterization and Analysis

Considerations of survivability and resiliency have always been of importance in the design and analysis of military systems. Over the p ast two decades, the importance of survivability and resiliency has expanded beyond mi litary systems to include public networks and infrastructure systems. The analysis a nd assessment of networked systems with respect to survivability has become particular ly acute in recent years, as attested to by a growing technical literature on the subject. In this paper, we bring these considerations of su rvivability and resiliency to bear on spacecraft and space-based networks. We develop a framework for comparing the survivability and resiliency of different space arc hitectures, namely that of a monolithic design and a distributed (or networked) space syste m architecture. There are multiple metrics along which different space architectures c an be benchmarked and compared. We argue that if survivability and resiliency are not accounted for, then the evaluation process is likely to be biased in favor of monolithic spacecra ft. We show that if in a given context survivability and resiliency are an important requi rement for a particular customer, then a distributed architecture is more likely to satisfy this requirement than a monolithic spacecraft design. We discuss in the context of our framework differe nt classes of threats, as well as the high-frequency and low-frequency system response to (or coping strategies with) these shocks or damaging events. We illustrate the import ance of this characterization for a formal definition of survivability and resiliency a nd a proper quantitative analysis of the subject. Finally, we propose in future work to inte grate our framework with a design tool that allows the exploration of the design trade-spa ce of distributed space architecture and show how survivability can be “optimized” or traded against other system attributes.

Satellite Electrical Power Subsystem: Statistical Analysis of On-Orbit Anomalies and Failures

A statistical analysis of satellite Electrical Power Subsystem (EPS) failures is conducted in this paper, and of particular interest is the comparison of the failures between the EPS of LEO and GEO satellites. 12To investigate in more details the failure mechanisms of the EPS, we also analyze the failure data of four subsystems of the EPS: the Battery / Cell subsystem, the Electrical Distribution subsystem (ED), the Solar Array Deployment subsystem (SAD), and the Solar Array Operating subsystem (SAO). Also, we categorized the anomaly and failure events into three groups based on the severity of the events: Minor, Major, and Fatal. The results show, first, that the EPS of GEO satellites experienced significantly more anomaly and failure events than the ones in LEO satellites; second, that the SAO subsystem was responsible for the majority of the anomaly and failure events experienced by the EPS of GEO satellites, indicating the vulnerability of that subsystem in GEO satellites, whereas the anomaly and failure events were spread amongst the EPS subsystems of LEO satellites; and third, that events in the EPS of GEO satellites tend to be less severe than in the EPS of LEO satellites. We hope that the results presented here would be useful to the space industry, for example, in redesigning test and screening programs for subsystems, or providing an empirical basis for redundancy allocation.

Stochastic Petri Nets for System Survivability and Multi- State Failure Analyses with Application to Space Sy stems

In this work, we explore the applicability of Stoch astic Petri Nets (SPNs) to multi-state failure and survivability analyses, using space sys tems as examples. Multi-state failure analyses introduce degraded states, and thus provid e more insights than the traditional binary reliability analysis into the degradation be havior of a system and its progression towards complete failure. Survivability analysis fo cuses on, among other things, the failure propagation in a system or a network following node or component failure, and it assesses for example whether the system will experience graceful degradation or catastrophic failure. The potential complexity of multi-state failure and system survivability analyses requires powerful (and flexible) stochastic modeling and simulation tools. After a brief introduction to Petri Nets, we argue that SPNs are particularly well suited to these types of analyses, and they are better at the task than the commonly used Markov Chains. We then propose a general framework for system survivability analysis, and we illustrate its applicability using space systems examples. We compare the survivability of two space architectures, a monolith spacecraft and a space-based network (which allows for distributed redundancy of certain subsystems). Monte Carlo simulations are run to generate representative results of stochastic behavior of the two architectures with r espect to on-orbit anomalies and failures. Finally, a comparison of the outputs of the two SPN models indicate, and quantify, the survivability advantage of the space-based network over the monolith spacecraft with respect to on-orbit anomalies and failures.